Bibliography

Bellon, A., and I. Zawadzki.

Forecasting of hourly accumulations of precipitation by optimal extrapolation of radar maps.
Journal of Hydrology, 157:-233, 1994.

Browning, K.A.

The FRONTIERS plan: A strategy for using radar and satellite imagery for very-short-range precipitation forecasting.
The Meteorological Magazine, 108:-184, 1979.

Browning, K.A., C.G. Collier, P.R. Larke, P. Menmuir, G.A. Monk, and R.G. Owens.

On the forecasting of frontal rain using a weather radar network.
Monthly Weather Review, 110:-552, 1982.

Cooley, J.W., and J.W. Tukey.

An algorithm for the machine calculation of complex fourier series.
Mathematical Computing, 19:-301, 1965.

Frigo, M., and S. Johnson.

FFTW: An adaptive software architecture for the FFT.
In International Conference on Acoustics, Speech and Signal Processing, pages 1381-1384, Seattle, WA, 1998.

Kreyszig, E.

Advanced Engineering Mathematics.
John Wiley and Sons, Inc., 1983.

Wolfson, M., B.E. Forman, R.G. Hallowell, and M.P. Moore.

The growth and decay storm tracker.
In 8th Conference on Aviation, pages 58-62, Dallas, TX, 1999. Amer. Meteor. Soc.

Figure 1:

Figure from Wolfson et al. (1999). The $5\times21$ pixel elliptical filter is shown at two of the eighteen possible orientations. The region of support for the filter is shown by the cross-hatched pixels.

Figure 2:

Original radar image, before large scale filtering. This is the lowest elevation of a volume scan by the Weather Service Radar, KFWS, in Fort Worth, Texas on May 8, 1995. The polar data from the radar were mapped into a Cartesian grid with each pixel being approximately $1km\times1km$ and clipped at a range of $256km$ from the radar.

Figure 3:

The effect of filtering the radar image in Figure 2 using the large scale filtering technique described in Wolfson et al. (1999). A $5\times21$ filter was used.

Figure 4:

The effect of filtering the radar image in Figure 2 using the faster large scale filtering technique described in this paper. Compare with the result of the original filtering technique (Figure 3). A $5\times21$ filter was used. The absolute difference between this image and Figure 3 is shown in Figure 5.

Figure 5:

The absolute difference between the results of filtering the radar image in Figure 2 using the faster large scale filtering technique described in this paper and using the original filtering technique (Figure 3). A $5\times21$ filter was used in both cases. The reasons for the difference in the outputs are explained in Section 2.6.

Table 1:

Image No.	$5\times21$ filters		$15\times64$ filters
	Original	Modified	Original	Modified
1	44	22	492	22
2	45	15	506	14
3	46	14	520	14
4	47	14	529	15
5	47	15	532	14
6	48	14	541	14
7	49	14	548	14

The time (in seconds) taken to filter each $512\times512$ image in a sequence using the original and modified filter technique for different filter sizes. Note that the modified technique is significantly faster. The modified technique's time requirement is also independent of filter sizes.